Multiuser scheduling via dynamic optimization

Abstract

This paper studies optimal scheduling in a multiuser wireless network, allowing for simultaneous transmission of sets of users. The approach is based on dynamic optimization. The multiuser scheduling problem is solved numerically as a dynamic programming problem with user-specific power constraints, assuming deterministic channel information during a given scheduling window. The dynamic programming approach enables to solve for both an optimal discrete power allocation and scheduling. Numerical examples suggest that previous results on optimal scheduling based on decomposing a T-period scheduling problem to T separate problems do not hold in general when decomposing the problem to a smaller number of larger problems. An auction-based algorithm for distributed scheduling is proposed, achieving an optimal schedule, assuming one-at-a-time transmission. © 2010 Springer-Verlag Berlin Heidelberg.